[Bio AI] 9월 세미나

Description

Bio AI의 9월 세미나 안내입니다.

이번 달에는 NVIDIA의 조예린 박사님과 Stanford의 이영은 박사님을 모시고, AI를 활용한 단백질 디자인과 뇌 기능 연구를 소개합니다.

• 일시 : 9월 19일(토), 12:30-3:30 PM
• 장소 : Dream Center in Sunnyvale [주소:1290 Reamwood Ave # B, Sunnyvale, CA 94089, 지도]
• RSVP  – 여기서
• 참석인원 : 최대 50명
• 문의사항:  BioAI 운영진- bio_ai@bayareakgroup.org
• RSVP의 컨펌 메일은 따로 발송되지 않습니다 (참석하시면 됩니다).
• 인원 제한을 넘어설 경우 혹은 일정 변경의 경우 등록해 주신 이메일 연락처로 공지됩니다. 또한 등록 시 사용하신 정보는 외부로 공유되지 않습니다.

Agenda

• 12:30 PM: Registration
• 1 – 3 PM: Seminar
• 3:30 PM: Close

[​​연사 1] 조예린 박사 (Yehlin Cho), Applied Research Scientist @ NVIDIA Digital Biology, Linkedin 

​Title: Joint Modeling of Protein Sequence and Structure for Design and Stability Prediction

Abstract: Most protein design approaches separate sequence and structure optimization. This talk presents joint sequence–structure modeling for all atom binder design through structure prediction model inversion, diffusion based generative modeling, and iterative sequence–structure refinement. Experiments on over 20,000 designed proteins demonstrate substantially greater stability from joint optimization than optimizing either independently. Building on these findings, I developed structure aware language models for absolute protein stability prediction, trained on 1.8 million cDNA display proteolysis measurements of proteins primarily drawn from the MGnify metagenomic database. Together, this work advances a unified framework for protein design and stability prediction, with applications in medicine and biotechnology.

[​연사 2] 이영은 박사 (Young-Eun Lee), Postdoctoral Fellow @ Stanford, LinkedIn

​Title: AI for the Human Brain: Speech Reconstruction, Task Prediction, and Cognitive Mapping

Abstract: Brain activity contains structure that can help infer signals and functions that are not directly observed. In this talk, I will present three studies: voice reconstruction from imagined-speech EEG, individualized task-activation prediction from resting-state fMRI, and ongoing work on task manifolds for organizing and interpreting brain activity. Together, these studies explore how shared structure in brain signals can support meaningful inference across different contexts.